WO2021166104A1

WO2021166104A1 - Processing device, estimation device, processing method, and estimation method

Info

Publication number: WO2021166104A1
Application number: PCT/JP2020/006445
Authority: WO
Inventors: 一峰小倉; 正行有吉
Original assignee: 日本電気株式会社
Priority date: 2020-02-19
Filing date: 2020-02-19
Publication date: 2021-08-26
Also published as: JPWO2021166104A1; JP7396452B2; US20230057701A1

Abstract

The present invention provides a processing device (10) comprising: an electromagnetic wave transmission and reception unit (11) which emits electromagnetic waves from a transmitting antenna and receives reflected waves via a receiving antenna; a label determination image generation unit (12) which generates a label determination image on the basis of a received reflected wave signal; a learning image generation unit (13) that generates a learning image on the basis of a signal that is a part of the received reflected wave signal and is smaller than a signal used for generating the label determination image; a label determination unit (14) which determines a label on the basis of the label determination image; and a teacher data generation unit (15) that generates teacher data in which the learning image and the label are linked, and stores the teacher data in a teacher data storage unit (16).

Description

Processing equipment, estimation equipment, processing method and estimation method

The present invention relates to a processing device, an estimation device, a processing method, and an estimation method.

Non-Patent Document 1 describes, in order to enhance security, marking and tracking a suspicious person, detecting a suspicious object with a microwave radar, identifying a person by face recognition, and inspecting belongings with a transparent image. It discloses what to do.

Patent Document 1 discloses that passengers' belongings are inspected using various technologies such as metal detectors and X-ray images, the opening and closing of gates is controlled according to the inspection results, and the traveling direction of passengers is controlled. doing.

Patent Document 2 discloses that when an intruder is detected by a millimeter-wave sensor, the line of sight of the surveillance camera is directed to the detection position of the intruder, and a zoom image of the intruder is obtained.

Special table 2017-537399 JP-A-2005-45712

The present inventor has found the following problems in object detection based on images of electromagnetic waves (eg, microwaves, millimeter waves, terahertz waves, etc.).

First, in the technique of estimating an object included in an image based on an estimation model generated from a labeled image (teacher data), the present inventor presents an electromagnetic wave used for image generation within a range in which an estimation result with sufficient accuracy can be obtained. By reducing the signal (reducing the amount of data), the processing load on the computer can be reduced, the size of the sensor device can be reduced by reducing the transmitting and receiving antennas, the cost burden can be reduced, the irradiation time can be shortened by reducing the transmitting antenna, and motion blur can be suppressed. I came up with the technology to realize.

However, if the amount of data is reduced based on whether or not the computer can recognize the object contained in the image with sufficient accuracy, the sharpness of the image will be reduced to the extent that a person cannot recognize the object contained in the image. In this case, in the teacher data generation scene, it becomes difficult to manually label the image by "a person browses the image, recognizes an object included in the image, and labels the image based on the recognition result".

The present invention is used in image generation within a range in which an estimation result with sufficient accuracy can be obtained in a technique for estimating an object included in an image based on an estimation model generated by a labeled image (teacher data) based on an electromagnetic signal. The challenge is to improve the difficulty of labeling work caused by reducing the signal of electromagnetic waves (reducing the amount of data).

According to the present invention
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
A processing device having the above is provided.

Further, according to the present invention.
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
An estimation model storage means for storing the estimation model generated by the processing apparatus having the
An estimation electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the estimation transmission antenna and receives reflected waves at the estimation reception antenna.
An estimation image generation means that generates an estimation image based on the received signal of the reflected wave, and
An estimation means for estimating an object included in the estimation image based on the estimation image and the estimation model, and
An estimation device having the above is provided.

Further, according to the present invention.
The computer
Irradiate electromagnetic waves from the transmitting antenna, receive reflected waves at the receiving antenna,
Based on the received signal of the reflected wave, a label determination image is generated.
A learning image is generated based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
The label is determined based on the label determination image, and the label is determined.
A processing method is provided for generating teacher data in which the learning image and the label are associated with each other and storing the teacher data in the teacher data storage means.

Further, according to the present invention.
The computer
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
The estimation model generated by the processing apparatus having the above is stored.
Electromagnetic waves are emitted from the estimation transmitting antenna, and the reflected wave is received by the estimation receiving antenna.
An estimation image is generated based on the received signal of the reflected wave, and the image is generated.
An estimation method for estimating an object included in the estimation image is provided based on the estimation image and the estimation model.

According to the present invention, in a technique for estimating an object included in an image based on an estimation model generated by a labeled image (teacher data) based on an electromagnetic signal, an image is generated within a range in which an estimation result with sufficient accuracy can be obtained. The difficulty of labeling work caused by reducing the signal of the electromagnetic wave used for the above (reducing the amount of data) is improved.

It is a figure which shows an example of the structure of the electromagnetic wave transmission / reception part for estimation of this embodiment. It is a figure which shows an example of the structure of the electromagnetic wave transmission / reception part for estimation of this embodiment. It is a figure which shows an example of the image generated by the estimation apparatus of this embodiment. It is a figure which shows an example of the structure of the electromagnetic wave transmission / reception part of this embodiment. It is a figure which shows an example of the image generated by the processing apparatus of this embodiment. It is a figure which shows an example of the hardware composition of the processing apparatus and estimation apparatus of this embodiment. This is an example of a functional block diagram of the estimation device of the present embodiment. It is a figure which shows an example of how to arrange the transmitting antenna and the receiving antenna of this embodiment. It is a figure which shows an example of the method of irradiating the electromagnetic wave from the plurality of transmitting antennas of this embodiment. It is a flowchart which shows an example of the processing flow of the estimation apparatus of this embodiment. It is an example of the functional block diagram of the processing apparatus of this embodiment. It is a figure which shows an example of how to arrange the transmitting antenna and the receiving antenna of this embodiment. It is a figure which shows an example of how to arrange the transmitting antenna and the receiving antenna of this embodiment. It is a figure which shows an example of the structure of the electromagnetic wave transmission / reception part of this embodiment. It is a figure which shows an example of the structure of the electromagnetic wave transmission / reception part for estimation of this embodiment. It is a flowchart which shows an example of the processing flow of the processing apparatus of this embodiment. It is a flowchart which shows an example of the processing flow of the estimation apparatus of this embodiment.

<First Embodiment>
"Overview of estimation system"
First, an outline of the estimation system of the present embodiment will be described. The estimation system of the present embodiment is an estimation device for inspecting the belongings of a person at an arbitrary place such as an airport or a station, and a processing device for generating an estimation model used in the inspection of the belongings by machine learning. And include.

For example, the estimation electromagnetic wave transmission / reception unit 21 (a part of the estimation device) as shown in FIG. 1 is installed at an arbitrary place such as an airport. The estimation electromagnetic wave transmission / reception unit 21 includes a transmission antenna for transmitting electromagnetic waves and a reception antenna for receiving electromagnetic waves. The estimation electromagnetic wave transmission / reception unit 21 irradiates the electromagnetic wave toward the person 1 passing through the predetermined position and receives the reflected wave. The estimation device generates an image based on the signal of the reflected wave, and estimates the possession of the person 1 based on the generated image and the estimation model generated in advance. In this way, according to the estimation device, it is possible to perform a walk-through type personal belongings inspection on a person 1 passing through a predetermined position.

FIG. 2 shows an example of a top view of the estimation electromagnetic wave transmission / reception unit 21. Then, FIG. 3 shows an example of an image generated based on the signal of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21. The frame portion of FIG. 3 is the location of the pistol (possession 2) possessed by the person 1 of FIG. It is difficult for the human eye to recognize the pistol in the frame portion of FIG. However, the computer can recognize the pistol in the frame portion with sufficient accuracy from the image shown in FIG. That is, the image generated based on the signal of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21 is so clear that the computer can recognize the object in the image with sufficient accuracy, but the object in the image cannot be recognized by a person. Can be a degree. In other words, the estimation electromagnetic wave transmission / reception unit 21 has a configuration capable of generating such an image.

Next, FIG. 4 shows an example of a top view of the electromagnetic wave transmission / reception unit 11 used for generating teacher data (labeled image) in the scene of generating the estimation model. The electromagnetic wave transmission / reception unit 11 includes a transmission antenna for transmitting electromagnetic waves and a reception antenna for receiving electromagnetic waves. Comparing FIG. 4 and FIG. 2, the electromagnetic wave transmission / reception unit 11 is different from the estimation electromagnetic wave transmission / reception unit 21 in that it has an additional portion 3. The presence of the transmitting antenna and the receiving antenna in the additional portion 3 makes it possible to receive the reflected wave of the possession 2 without leaking it and generate an image showing the possession 2 more clearly.

FIG. 5 shows an example of an image generated based on the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11. The frame portion of FIG. 5 is the location of the pistol (possession 2) possessed by the person 1 of FIG. In the case of this example, even the human eye can recognize the pistol in the frame portion of FIG. 5 from its shape. Naturally, the computer can recognize the pistol in the frame portion with sufficient accuracy from the image shown in FIG.

The processing device 10 generates a label determination image and a learning image based on the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11.

Specifically, the processing device 10 generates a label determination image by using all or most of the reflected wave signals received by the electromagnetic wave transmission / reception unit 11. As shown in FIG. 5, the label determination image generated in this manner is an image with a sharpness that allows a person to recognize an object in the image.

Further, the processing device 10 generates a learning image by using a part of the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11. As shown in FIG. 4, the learning image generated in this way has a sharpness that the computer can recognize the object in the image with sufficient accuracy, but the object in the image cannot be recognized by a person. That is, the learning image has the same sharpness as the image used in the estimation process by the estimation device 20.

Then, when the processing device 10 determines the label based on the label determination image, the processing device 10 generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the storage device.

As described above, the estimation system of the present embodiment is an amount of data used for generating an image used in the process of estimating the belongings of the person 1 (data used by the estimation device 20 for generating an image used in the estimation process). The amount, the amount of data used to generate the image to be the teacher data) is sufficiently reduced within the range where the estimation result with sufficient accuracy by the computer can be obtained. As a result, the processing load of the computer can be reduced, the sensor device can be downsized and the cost burden can be reduced by reducing the transmitting and receiving antennas, the irradiation time can be shortened by reducing the transmitting antenna, and the motion blur can be suppressed.

The estimation system of the present embodiment has a configuration capable of acquiring more reflected wave data than the estimation electromagnetic wave transmitter / receiver 21 actually installed and used at an airport or the like in the scene of generating the estimation model. The electromagnetic wave transmission / reception unit 11 is used to irradiate an object with an electromagnetic wave and receive a reflected wave. Then, the estimation system of the present embodiment is based on the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11, an image for determining a label having a sharpness that allows a person to recognize an object in the image, and an estimation by the estimation device 20. A learning image with the same degree of sharpness as the image used in the processing is generated. Then, the label determined based on the label determination image and the learning image are associated with each other to generate teacher data. With such a configuration, the labeling work can be performed without any problem even if the amount of data used for generating the image used in the process of estimating the possession of the person 1 is reduced as described above.

"Configuration of estimation device"
Next, the configuration of the estimation device will be described in detail. First, an example of the hardware configuration of the estimation device will be described. Each functional unit included in the estimation device is stored in the CPU (Central Processing Unit) of an arbitrary computer, memory, a program loaded in the memory, and a storage unit such as a hard disk for storing the program (stored from the stage of shipping the device in advance). In addition to the existing programs, it can also store programs downloaded from storage media such as CDs (Compact Discs) and servers on the Internet), and it is realized by any combination of hardware and software centered on the network connection interface. NS. And, it is understood by those skilled in the art that there are various modifications of the realization method and the device.

FIG. 6 is a block diagram illustrating the hardware configuration of the estimation device. As shown in FIG. 6, the estimation device includes a processor 1A, a memory 2A, an input / output interface 3A, a peripheral circuit 4A, a bus 5A, and an electromagnetic wave transmission / reception device 6A. The peripheral circuit 4A includes various modules. The processing system 20 does not have to have the peripheral circuit 4A. The processing system 20 may be composed of a plurality of physically and / or logically separated devices, or may be composed of one physically and / or logically integrated device. When the processing system 20 is composed of a plurality of physically and / or logically separated devices, each of the plurality of devices may have the above hardware configuration.

The bus 5A is a data transmission path for the processor 1A, the memory 2A, the peripheral circuits 4A, and the input / output interface 3A to send and receive data to and from each other. The processor 1A is, for example, an arithmetic processing unit such as a CPU or a GPU (Graphics Processing Unit). The memory 2A is, for example, a memory such as a RAM (RandomAccessMemory) or a ROM (ReadOnlyMemory). The input / output interface 3A includes an interface for acquiring information from an input device, an external device, an external server, an external sensor, a camera, etc., an interface for outputting information to an output device, an external device, an external server, etc. .. As an example of the external sensor, the electromagnetic wave transmission / reception device 6A is shown. The electromagnetic wave transmitting / receiving device 6A includes a transmitting antenna for transmitting electromagnetic waves and a receiving antenna for receiving electromagnetic waves. The electromagnetic wave transmission / reception device 6A is, for example, a radar. The input device is, for example, a keyboard, a mouse, a microphone, a physical button, a touch panel, or the like. The output device is, for example, a display, a speaker, a printer, a mailer, or the like. The processor 1A can issue commands to each module and perform calculations based on the calculation results thereof.

Next, the functional configuration of the estimation device will be described. FIG. 7 shows an example of a functional block diagram of the estimation device 20. As shown in the figure, the estimation device 20 includes an estimation electromagnetic wave transmission / reception unit 21, an estimation image generation unit 22, an estimation unit 23, an estimation model storage unit 24, and an estimation parameter setting unit 25.

The estimation model storage unit 24 stores the estimation model generated by the processing device 10. The details of the estimation model will be clarified in the following description of the configuration of the processing device 10.

The estimation electromagnetic wave transmission / reception unit 21 includes an estimation transmission antenna and an estimation reception antenna. Then, the estimation electromagnetic wave transmission / reception unit 21 irradiates the electromagnetic wave from the estimation transmission antenna and receives the reflected wave at the estimation reception antenna. The estimation electromagnetic wave transmission / reception unit 21 is, for example, a radar. The electromagnetic wave transmitted and received by the estimation electromagnetic wave transmission / reception unit 21 is, for example, an electromagnetic wave having a wavelength of 30 micrometers or more and 1 meter or less (example: microwave, millimeter wave, terahertz wave, etc.). The estimation electromagnetic wave transmission / reception unit 21 can be configured by adopting any technique. For example, the estimation electromagnetic wave transmission / reception unit 21 may be a sensor panel in which a plurality of estimation transmission antennas and a plurality of estimation reception antennas are arranged as in the example of FIG. The plurality of estimation transmitting antennas irradiate electromagnetic waves in a predetermined order with their timings shifted from each other. Then, the reflected wave is received by all of the plurality of estimation receiving antennas.

In the example shown in FIG. 1, two sensor panels facing each other constitute the estimation electromagnetic wave transmission / reception unit 21, but one sensor panel may form the estimation electromagnetic wave transmission / reception unit 21. The estimation electromagnetic wave transmission / reception unit 21 may be configured by one or more sensor panels. Further, in the example shown in FIG. 1, a gate is created by two sensor panels so that a person 1 can pass between them. However, for example, the sensor panel is embedded in a wall or the like, and the person cannot recognize the existence of the sensor panel. You may do so.

Returning to FIG. 7, the estimation parameter setting unit 25 sets various parameters related to the transmission / reception of electromagnetic waves by the estimation electromagnetic wave transmission / reception unit 21. For example, the estimation parameter setting unit 25 sets the irradiation order of the plurality of estimation transmission antennas, the frequency of the electromagnetic wave emitted by each estimation transmission antenna, the irradiation time of each estimation transmission antenna, and the like. For example, as shown in FIG. 9, the frequencies of the electromagnetic waves emitted from the plurality of estimation transmitting antennas can be changed according to the time. The estimation parameter setting unit 25 can set the various parameters based on the user input. The estimation electromagnetic wave transmission / reception unit 21 transmits / receives electromagnetic waves based on various parameters set by the estimation parameter setting unit 25.

The estimation image generation unit 22 generates an estimation image based on the signal of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21. The estimation image generation unit 22 can generate an estimation image based on, for example, the following equation (1).

The estimation image is three-dimensional and is composed of a set of values of each of a plurality of voxels. The v vector indicates the center position of one voxel. The R _m vector indicates the position of the transmitting antenna (estimating transmitting antenna). The R _n vector indicates the position of the receiving antenna (estimating receiving antenna). s _{m, n, and f} indicate signals of electromagnetic waves (reflected waves) having a frequency f that the transmitting antenna m irradiates and the receiving antenna n receives. In F (f), j indicating the frequency at the time index f is an imaginary number, and c is the speed of light.

FIG. 3 shows an example of the estimation image generated by the estimation image generation unit 22. The frame portion of FIG. 3 is the location of the pistol. As shown in the figure, the estimation image can be sharp enough that the computer can recognize the object in the image with sufficient accuracy, but the object in the image cannot be recognized by a person. In other words, the estimation electromagnetic wave transmission / reception unit 21 has a configuration in which such an image is generated (the number of estimation transmission antennas, the number of estimation reception antennas, the arrangement method, the frequency used and the number thereof, etc.). ing.

The estimation unit 23 estimates an object included in the estimation image based on the estimation image generated by the estimation image generation unit 22 and the estimation model stored in the estimation model storage unit 24.

Next, an example of the processing flow of the estimation device 20 will be described with reference to the flowchart of FIG.

When the processing is started, the estimation electromagnetic wave transmission / reception unit 21 repeats irradiation of electromagnetic waves having a wavelength of 30 micrometers or more and 1 meter or less and reception of reflected waves at predetermined intervals. Then, when the estimation image generation unit 22 acquires the signal of the reflected wave output by the estimation electromagnetic wave transmission / reception unit 21 (S10), the estimation image generation unit 22 generates an estimation image based on the signal (S11).

Next, the estimation unit 23 estimates the object included in the estimation image based on the estimation image generated by the estimation image generation unit 22 in S11 and the estimation model stored in the estimation model storage unit 24 in advance. (S12).

Then, the estimation device 20 outputs the estimation result (S13). For example, the estimation device 20 may output the estimation result via an arbitrary output device such as a display, a projection device, a speaker, a printer, and a mailer. The estimation result includes information (name, etc.) about the object estimated to be included in the estimation image.

In addition, the estimation device 20 may perform warning processing when it is estimated that a predetermined object is included in the estimation image. The warning process includes, but is not limited to, lighting of a warning lamp, output of a warning sound, output of warning information via an output device such as a display, and the like. Predetermined objects are objects that are not allowed to be carried at that location, such as dangerous goods such as pistols and knives.

"Processing device configuration"
The configuration of the processing device will be described in detail. An example of the hardware configuration of the processing device is the same as the example of the hardware configuration of the estimation device 20 described above.

Next, the functional configuration of the processing device will be described. FIG. 11 shows an example of a functional block diagram of the processing device 10. As shown in the figure, the processing device 10 includes an electromagnetic wave transmission / reception unit 11, a label determination image generation unit 12, a learning image generation unit 13, a label determination unit 14, a teacher data generation unit 15, and a teacher data storage unit. It has 16, an estimation model generation unit 17, and a parameter setting unit 18.

The electromagnetic wave transmission / reception unit 11 includes a transmission antenna and a reception antenna. Then, the electromagnetic wave transmitting / receiving unit 11 irradiates the electromagnetic wave from the transmitting antenna and receives the reflected wave at the receiving antenna. The electromagnetic wave transmission / reception unit 11 is, for example, a radar. The electromagnetic wave transmitted and received by the electromagnetic wave transmission / reception unit 11 is, for example, an electromagnetic wave having a wavelength of 30 micrometers or more and 1 meter or less (example: microwave, millimeter wave, terahertz wave, etc.). The electromagnetic wave transmission / reception unit 11 can be configured by adopting any technique. For example, the electromagnetic wave transmitting / receiving unit 11 may be a sensor panel in which a plurality of transmitting antennas and a plurality of receiving antennas are arranged as in the example of FIG. The plurality of transmitting antennas irradiate electromagnetic waves in a predetermined order at different timings. Then, the reflected wave is received by all of the plurality of receiving antennas.

The electromagnetic wave transmission / reception unit 11 is configured to receive the reflected wave of the possession 2 without leaking more than the estimation electromagnetic wave transmission / reception unit 21.

For example, the electromagnetic wave transmission / reception unit 11 may include more transmission antennas and reception antennas than the estimation electromagnetic wave transmission / reception unit 21. That is, the number of estimation transmitting antennas of the estimation electromagnetic wave transmission / reception unit 21 is smaller than the number of transmission antennas of the electromagnetic wave transmission / reception unit 11, and the number of estimation reception antennas of the estimation electromagnetic wave transmission / reception unit 21 is the reception of the electromagnetic wave transmission / reception unit 11. It may be less than the number of antennas. If the number of transmitting antennas (estimating transmitting antennas) and receiving antennas (estimating receiving antennas) is small, the number of reflected wave signals that can be acquired simply decreases (the amount of data decreases). As a result, the sharpness of the image generated from the signal can be reduced. Further, the number of frequencies used by the estimation electromagnetic wave transmission / reception unit 21 may be smaller than the number of frequencies used by the electromagnetic wave transmission / reception unit 11. Reducing the number of frequencies used can reduce the sharpness of the resulting image.

Further, the electromagnetic wave transmission / reception unit 11 may have a longer opening length than the estimation electromagnetic wave transmission / reception unit 21. The aperture length is the distance between the transmitting antennas (estimating transmitting antennas) located at both ends in the left-right direction among the plurality of transmitting antennas (estimating transmitting antennas). Wt in FIG. 12 indicates the opening length. The aperture length is the distance between the receiving antennas (estimated receiving antennas) located at both ends in the left-right direction among the plurality of receiving antennas (estimated receiving antennas). Wr in FIG. 12 indicates the opening length.

Note that the aperture length may be a concept that includes the distance between the transmitting antennas (estimating transmitting antennas) located at both ends in the vertical direction among the plurality of transmitting antennas (estimating transmitting antennas). Ht in FIG. 12 indicates the opening length. Further, the aperture length may be a concept including the distance between the receiving antennas (estimated receiving antennas) located at both ends in the vertical direction among the plurality of receiving antennas (estimated receiving antennas). Hr in FIG. 12 indicates the opening length.

If the opening length is short, reception leakage of reflected waves may occur. For example, if the belongings 2 are tilted with respect to the radar surface as shown in FIG. 2, reception omission of the reflected wave (arrow in FIG. 2) reflected by the belongings 2 may occur. On the other hand, as shown in FIG. 4, when the opening length is long, the reflected wave reflected by the possession 2 is reflected by the possession 2 even when the possession 2 is tilted with respect to the radar surface as shown in FIG. (Arrow in FIG. 4) can be received.

Further, although the electromagnetic wave transmission / reception unit 11 has more transmission antennas and reception antennas than the estimation electromagnetic wave transmission / reception unit 21, even if the electromagnetic wave transmission / reception unit 11 and the estimation electromagnetic wave transmission / reception unit 21 have the same opening length. good. An example of this is shown in FIGS. 12 and 13. FIG. 12 is a configuration example of the electromagnetic wave transmission / reception unit 11, and FIG. 13 is a configuration example of the estimation electromagnetic wave transmission / reception unit 21. In the case of the illustrated example, the electromagnetic wave transmission / reception unit 11 includes more transmission antennas and reception antennas than the estimation electromagnetic wave transmission / reception unit 21. However, the opening lengths of the electromagnetic wave transmission / reception unit 11 and the estimation electromagnetic wave transmission / reception unit 21 are about the same.

12 and 13 have more transmitting antennas and receiving antennas than the above-mentioned "estimating electromagnetic wave transmitting / receiving unit 21" by devising the arrangement of the transmitting antenna and the receiving antenna in one sensor panel. The opening lengths of the transmission / reception unit 11 and the estimation electromagnetic wave transmission / reception unit 21 are the same. As another example, as shown in FIGS. 15 and 16, by devising the arrangement of the transmitting antenna and the receiving antenna in the entire plurality of sensor panels, the above-mentioned “more transmitting antennas than the estimation electromagnetic wave transmitting / receiving unit 21” And the receiving antenna is provided, but the above-mentioned opening lengths of the electromagnetic wave transmission / reception unit 11 and the estimation electromagnetic wave transmission / reception unit 21 may be the same.

Note that, as in the examples of FIGS. 2 and 4, the configuration of the electromagnetic wave transmission / reception unit 11 may be a configuration in which a partial configuration (additional portion 3) is added to the configuration of the estimation electromagnetic wave transmission / reception unit 21. In this way, the reflected wave signal received by the electromagnetic wave transmitting / receiving unit 11 is generally "a reflected wave signal acquired when the configuration of the estimation electromagnetic wave transmitting / receiving unit 21 is adopted" and "a part of the addition". It can be considered as a combination with "the signal of the reflected wave acquired by the configuration". In this case, for example, the above-mentioned learning is performed based on the "reflected wave signal acquired when the configuration of the estimation electromagnetic wave transmission / reception unit 21 is adopted" by removing the "reflected wave signal acquired by the added partial configuration". By generating the image for estimation, it is possible to generate the estimation model by machine learning based on the image (teacher data) generated under almost the same conditions as the image used at the time of estimation (example: completely under the same conditions). It becomes. That is, the data acquisition environment (number and position of transmitting / receiving antennas, frequency used and the number thereof, etc.) for learning image generation by the learning image generation unit 13 and the estimation image generation by the estimation image generation unit 22. It is possible to completely match the data acquisition environment (number and position of transmitting / receiving antennas, frequency used and the number thereof, etc.). As a result, improvement of estimation accuracy is expected.

Returning to FIG. 11, the parameter setting unit 18 sets various parameters related to the transmission / reception of electromagnetic waves by the electromagnetic wave transmission / reception unit 11. For example, the parameter setting unit 18 sets the irradiation order of the plurality of estimation transmitting antennas, the frequency of the electromagnetic wave emitted by each estimation transmitting antenna, the irradiation time of each estimation transmitting antenna, and the like. For example, as shown in FIG. 9, the frequencies of electromagnetic waves emitted from a plurality of transmitting antennas can be changed according to time. The parameter setting unit 18 can set the various parameters based on the user input. The electromagnetic wave transmission / reception unit 11 transmits / receives electromagnetic waves based on various parameters set by the parameter setting unit 18.

Returning to FIG. 11, the label determination image generation unit 12 generates a label determination image based on the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11. The process of generating an image based on the received reflected wave signal is the same as the image generation process described in the estimation image generation unit 22. The label determination image generation unit 12 generates a label determination image by using all or most of the signals of the reflected wave received by the electromagnetic wave transmission / reception unit 11. As shown in FIG. 5, the label determination image generated in this manner is an image with a sharpness that allows a person to recognize an object in the image.

Returning to FIG. 11, the learning image generation unit 13 is a part of the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11, and is used for generating the label determination image by the label determination image generation unit 12. A training image is generated based on a signal that is less than the signal to be generated. In the learning image generated in this way, the object in the image can be recognized with sufficient accuracy by the computer, but the sharpness can be such that the object in the image cannot be recognized by a person. That is, the learning image can have the same degree of sharpness as the image generated by the estimation image generation unit 22 described above. The process of generating an image based on the received reflected wave signal is the same as the image generation process described in the estimation image generation unit 22.

Here, a method of selecting a part to be used for generating a learning image from the signals of the reflected wave received by the electromagnetic wave transmission / reception unit 11 will be described.

For example, the learning image generation unit 13 may randomly select a part to be used for learning image generation from the signals of the reflected waves received by the electromagnetic wave transmission / reception unit 11. The number of data to be selected is preferably the same as the number of data used to generate an image by the estimation image generation unit 22.

As another example, the learning image generation unit 13 selects a part from a plurality of transmitting antennas, and the electromagnetic wave emitted by the selected transmitting antenna from the reflected wave signals received by the electromagnetic wave transmitting / receiving unit 11. The signal excluding the signal of the reflected wave of the above may be selected as a part used for generating the image for learning.

Here, an example of a method of selecting a part of the transmitting antennas will be described. For example, as in the examples of FIGS. 2 and 4, when the configuration of the electromagnetic wave transmission / reception unit 11 is a configuration in which a partial configuration (additional portion 3) is added to the configuration of the estimation electromagnetic wave transmission / reception unit 21, a learning image is generated. The unit 13 may select the transmitting antenna of the additional unit 3. In addition, the learning image generation unit 13 may randomly select a predetermined number of transmitting antennas from the plurality of transmitting antennas. It is preferable that the position and / or number of the transmitting antennas remaining without being selected is the same at the position / number / or both of the estimation transmitting antennas included in the estimation electromagnetic wave transmission / reception unit 21.

As another example, the learning image generation unit 13 selects a part from a plurality of receiving antennas, and the reflection received by the selected receiving antenna from the reflected wave signals received by the electromagnetic wave transmitting / receiving unit 11. The signal excluding the wave signal may be selected as part of the training image generation.

Here, an example of a method of selecting a part of the receiving antennas will be described. For example, as in the examples of FIGS. 2 and 4, when the configuration of the electromagnetic wave transmission / reception unit 11 is a configuration in which a partial configuration (additional portion 3) is added to the configuration of the estimation electromagnetic wave transmission / reception unit 21, a learning image is generated. The unit 13 may select the receiving antenna of the additional unit 3. In addition, the learning image generation unit 13 may randomly select a predetermined number of receiving antennas from the plurality of receiving antennas. It is preferable that the position and / or number of the receiving antennas remaining without being selected is the same as the position / number / or both of the estimation receiving antennas included in the estimation electromagnetic wave transmission / reception unit 21.

As another example, the learning image generation unit 13 may select a part from a plurality of transmitting antennas and a part from a plurality of receiving antennas. Then, the learning image generation unit 13 receives the reflected wave signal of the electromagnetic wave irradiated by the selected transmitting antenna and the reflection received by the selected receiving antenna from the reflected wave signals received by the electromagnetic wave transmitting / receiving unit 11. The signal excluding the wave signal may be selected as a part to be used for learning image generation.

Here, an example of a method of selecting a part of the transmitting antenna and the receiving antenna will be described. For example, as in the examples of FIGS. 2 and 4, when the configuration of the electromagnetic wave transmission / reception unit 11 is a configuration in which a partial configuration (additional portion 3) is added to the configuration of the estimation electromagnetic wave transmission / reception unit 21, a learning image is generated. The unit 13 may select the transmitting antenna and the receiving antenna of the additional part 3. In addition, the learning image generation unit 13 may be selected as a part of randomly removing a predetermined number of transmitting antennas and receiving antennas from the plurality of transmitting antennas. The position and number of the transmitting antenna and the receiving antenna remaining without being selected, or both, are the same as the position and / or both of the estimation transmitting antenna and the estimation receiving antenna included in the estimation electromagnetic wave transmission / reception unit 21. Is preferable.

As another example, as described above, when the electromagnetic wave transmission / reception unit 11 irradiates the electromagnetic wave from the transmission antenna while changing the frequency, the learning image generation unit 13 receives the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11. A signal excluding the signal of the selected frequency may be selected as a part to be used for generating the learning image.

Further, the learning image generation unit 13 selects a part from the plurality of transmitting antennas, selects a part from the plurality of receiving antennas, and further selects a part from the frequencies to be used. May be good. Then, the learning image generation unit 13 receives the reflected wave signal of the electromagnetic wave irradiated by the selected transmitting antenna and the reflected wave received by the selected receiving antenna from the reflected wave signals received by the electromagnetic wave transmitting / receiving unit 11. The signal and the signal excluding the signal of the selected frequency may be selected as a part to be used for the image generation for learning.

Returning to FIG. 11, the label determination unit 14 determines the label based on the label determination image generated by the label determination image generation unit 12.

For example, the label determination unit 14 has a means for outputting a label determination image and a means for accepting user input of the label of the output label determination image (name and position of an object included in the label determination image, etc.). You may. The output of the label determination image is realized based on any image output device such as a display, a projection device, a printer, and a mailer. In addition, user input of labels is realized based on all input devices such as keyboards, mice, microphones, physical buttons, and touch panels, and arbitrary UI (user interface) screens.

As another example, the image from the image having the same sharpness as the label determination image by machine learning based on the labeled image (teacher data) having the same sharpness as the label determination image in advance. An estimation model for estimating the name and position of the object included in may be generated. Then, the label determination unit 14 estimates the object included in the label determination image based on the label determination image and the estimation model, and obtains the estimation result (name and position of the object, etc.) of the label determination image. It may be determined as a label of.

The teacher data generation unit 15 generates teacher data in which the learning image generated by the learning image generation unit 13 and the label determined by the label determination unit 14 are associated with each other, and stores the teacher data in the teacher data storage unit 16.

The estimation model generation unit 17 generates an estimation model that estimates an object included in the image from the image generated based on the signal of the reflected wave of the electromagnetic wave by machine learning based on the teacher data stored in the teacher data storage unit 16. do.

Next, an example of the processing flow of the processing apparatus 10 will be described with reference to the flowchart of FIG.

First, the electromagnetic wave transmission / reception unit 11 irradiates the electromagnetic wave from the transmitting antenna and receives the reflected wave at the receiving antenna (S20).

Next, the label determination unit 12 generates a label determination image based on the signal of the reflected wave received in S20 (S23). Next, the label determination unit 14 displays the label determination image on, for example, a display. (S24), the user input of the label of the displayed label determination image is accepted (S25).

Further, after S20, the learning image generation unit 13 selects a part of the reflected wave signal received in S20 to be used for learning image generation (S21). Then, the learning image generation unit 13 generates a learning image based on a part of the selected signals (S22). The signal (data) used to generate the learning image is smaller than the signal (data) used to generate the label determination image.

Next, the teacher data generation unit 15 generates teacher data in which the learning image generated in S22 and the label input in S25 are associated with each other, and stores the teacher data in the teacher data storage unit 16 (S26). After that, by repeating the same process, the teacher data is accumulated in the teacher data storage unit 16.

In the processing example of FIG. 14, each time the reflected wave signal is received, the learning image is generated (S21, S22), the label determination image is generated (S23), the label is determined (S24, S25), and so on. Then, a series of processing of generation / accumulation of teacher data (S26) was performed.

As a modification, a process of generating a learning image (S21, S22) and a label determination image (S23), a process of determining a label (S24, S25), and a process of generating / accumulating teacher data (S26). And may be separated and each may be performed independently.

That is, each time the reflected wave signal is received (S20), only the learning image is generated (S21, S22) and the label determination image is generated (S23), and the learning image and the label determination image are generated. Then, they may be linked and stored in the storage means. By repeating this process, the information associated with the learning image and the label determination image is accumulated in the storage means.

Then, at an arbitrary timing thereafter, labeling processing (label determination (S24, S25) and teacher data generation / accumulation (S26)) for a plurality of learning images stored in the storage means is performed by batch processing. It may be done.

"Effect of estimation system"
The estimation system described above includes the amount of data used to generate an image used in the process of estimating the belongings of person 1 (the amount of data used by the estimation device 20 to generate an image used in the estimation process, and teacher data. The amount of data used to generate the image to be used) should be sufficiently reduced to the extent that a computer can obtain a sufficiently accurate estimation result. As a result, the processing load of the computer can be reduced, the sensor device can be downsized and the cost burden can be reduced by reducing the transmitting and receiving antennas, the irradiation time can be shortened by reducing the transmitting antenna, and the motion blur can be suppressed.

In addition, the estimation system of the present embodiment can select a part of the reflected wave signal received by the electromagnetic wave transmission / reception unit 11 to be used for learning image generation by a characteristic method. For example, a part of the transmitting antenna and the receiving antenna can be selected, and the signal of the reflected wave due to the electromagnetic wave transmitted by the selected transmitting antenna and the signal of the reflected wave received by the selected receiving antenna can be removed. In this case, if a part of the transmitting antenna and the receiving antenna are appropriately selected, the signal used for learning image generation will be "the signal of the reflected wave acquired when the configuration of the estimation electromagnetic wave transmission / reception unit 21 is adopted". It will be similar. Then, the above-mentioned learning image can be generated based on the "signal of the reflected wave acquired when the configuration of the estimation electromagnetic wave transmission / reception unit 21 is adopted". As a result, it becomes possible to generate an estimation model by machine learning based on an image (teacher data) generated under almost the same conditions as the image used at the time of estimation, and improvement in estimation accuracy is expected.

<Second embodiment>
In the estimation system of this embodiment, the electromagnetic wave transmission / reception unit 11 and the estimation electromagnetic wave transmission / reception unit 21 have the same configuration. That is, the number of transmitting antennas constituting the electromagnetic wave transmission / reception unit 11 and the number of estimation transmission antennas constituting the estimation electromagnetic wave transmission / reception unit 21 are the same, and the arrangement method is the same. Further, the number of receiving antennas constituting the electromagnetic wave transmission / reception unit 11 and the number of estimation reception antennas constituting the estimation electromagnetic wave transmission / reception unit 21 are the same, and the arrangement method is the same.

The parameter setting unit 18 and the estimation parameter setting unit 25 of the present embodiment set the values of various parameters to the same values. That is, the irradiation order of each of the plurality of transmitting antennas is the same as that of the corresponding (same arrangement position) estimation transmitting antennas in the plurality of estimation transmitting antennas. Further, the frequency of the electromagnetic wave emitted by each of the plurality of transmitting antennas is the same as that of the corresponding (same arrangement position) estimation transmitting antennas in the plurality of estimation transmitting antennas. Further, the irradiation time of each of the plurality of transmitting antennas is the same as the irradiation time of the corresponding (same arrangement position) estimation transmitting antennas in the plurality of estimation transmitting antennas.

Then, the estimation image generation unit 22 generates an estimation image based on a part of the signals of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21.

Here, a method of selecting a part to be used for generating the estimation image from the signals of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21 will be described.

For example, the estimation image generation unit 22 may randomly select a signal to be used for generating the estimation image from the signals of the reflected waves received by the estimation electromagnetic wave transmission / reception unit 21.

Further, the estimation image generation unit 22 selects the signal of the reflected wave based on the electromagnetic wave emitted from the selected estimation transmission antenna and the signal of the reflected wave received by the estimation electromagnetic wave transmission / reception unit 21. A signal excluding at least one of the reflected wave signals received by the estimation receiving antenna may be selected as a signal to be used for generating the estimation image.

Further, the estimation image generation unit 22 selects a signal obtained by excluding the signal of the selected frequency from the reflected wave signals received by the estimation electromagnetic wave transmission / reception unit 21 as a signal to be used for generating the estimation image. You may.

These methods by the estimation image generation unit 22 are for generating a learning image from the signal of the reflected wave received by the electromagnetic wave transmission / reception unit 11 by the learning image generation unit 13 described in the first embodiment. It is the same as "How to select a part to be used for".

The estimation image generation unit 22 and the learning image generation unit 13 use the same method to obtain an estimation image and learning from the reflected wave signals received by the estimation electromagnetic wave transmission / reception unit 21 and the electromagnetic wave transmission / reception unit 11. It is preferable to select a part to be used for image generation.

When the processing is started, the estimation electromagnetic wave transmission / reception unit 21 repeats irradiation of electromagnetic waves having a wavelength of 30 micrometers or more and 1 meter or less and reception of reflected waves at predetermined intervals. Then, when the estimation image generation unit 22 acquires the signal of the reflected wave output by the estimation electromagnetic wave transmission / reception unit 21 (S20), the estimation image generation unit 22 selects a part to be used for the estimation image generation from the acquired signals (S20). S31). Then, the estimation image generation unit 22 generates an estimation image based on the selected signal (S32).

Next, the estimation unit 23 estimates the object included in the estimation image based on the estimation image generated by the estimation image generation unit 22 in S32 and the estimation model stored in the estimation model storage unit 24 in advance. (S33).

Then, the estimation device 20 outputs the estimation result (S34). For example, the estimation device 20 may output the estimation result via an arbitrary output device such as a display, a projection device, a speaker, a printer, and a mailer. The estimation result includes information (name, etc.) about the object estimated to be included in the estimation image.

Other configurations of the processing system of the present embodiment are the same as those of the processing system of the first embodiment.

According to the estimation system of the present embodiment described above, the estimation result with sufficient accuracy in the technique of estimating the object included in the image based on the estimation model generated by the labeled image (teacher data) based on the electromagnetic signal. The difficulty of the labeling work caused by reducing the signal of the electromagnetic wave used for image generation (reducing the amount of data) is improved within the range where the above can be obtained.

Further, according to the estimation system of the present embodiment, the processing system of the first embodiment realizes "miniaturization of the sensor device and reduction of cost burden by reducing the transmission / reception antennas, shortening of the irradiation time by reducing the transmission antennas, and the like. Although "suppression of motion blur, etc." cannot be realized, the amount of data used to generate an image used in the process of estimating the belongings of person 1 (the amount of data used by the estimation device 20 to generate an image used in the estimation process). , The amount of data used to generate images as teacher data) is reduced, thereby reducing the processing load on the computer.

Further, according to the estimation system of the present embodiment, the estimation image generation unit 22 and the learning image generation unit 13 use the same method to obtain the reflected waves received by the estimation electromagnetic wave transmission / reception unit 21 and the electromagnetic wave transmission / reception unit 11. From the signals, a part to be used for generating the estimation image and the learning image can be selected. In this case, it becomes possible to generate an estimation model by machine learning based on an image (teacher data) generated under substantially the same conditions as the image used at the time of estimation, and improvement in estimation accuracy is expected.

Although the invention of the present application has been described above with reference to the embodiments (and examples), the invention of the present application is not limited to the above-described embodiments (and examples). Various changes that can be understood by those skilled in the art can be made within the scope of the present invention in terms of the structure and details of the present invention.

Some or all of the above embodiments may also be described, but not limited to:
1. 1. Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
Processing equipment with.
2. The processing apparatus according to 1, further comprising an estimation model generation means for generating an estimation model by machine learning based on the teacher data stored in the teacher data storage means.
3. 3. The label determination image generation means
A means for outputting the label determination image and
A means for accepting the user input of the label of the output image for determining the label, and
The processing apparatus according to 1 or 2.
4. The processing device according to any one of 1 to 3, wherein the learning image generation means randomly selects a signal to be used for generating the learning image from the received signals of the reflected wave.
5. The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The learning image generation means receives the signal of the reflected wave based on the electromagnetic wave emitted from the selected transmitting antenna from the received signals of the reflected wave, and the signal received by the selected receiving antenna. The processing apparatus according to any one of 1 to 3, wherein a signal excluding at least one of the reflected wave signals is selected as a signal to be used for generating the learning image.
6. The electromagnetic wave transmitting / receiving means irradiates the electromagnetic wave from the transmitting antenna while changing the frequency.
The learning image generation means 1 to 3 and 5 select a signal obtained by excluding a signal having a selected frequency from the received signals of the reflected wave as a signal to be used for generating the learning image. The processing apparatus according to any one.
7. Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
An estimation model storage means for storing the estimation model generated by the processing apparatus having the
An estimation electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the estimation transmission antenna and receives reflected waves at the estimation reception antenna.
An estimation image generation means that generates an estimation image based on the received signal of the reflected wave, and
An estimation means for estimating an object included in the estimation image based on the estimation image and the estimation model, and
Estimator with.
8. The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The estimation electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of estimation transmitting antennas, receives reflected waves by the plurality of estimation receiving antennas, and receives reflected waves.
The electromagnetic wave transmitting / receiving means and the estimation electromagnetic wave transmitting / receiving means have the same configuration.
7. The estimation device according to 7, further comprising a parameter setting means for making the irradiation order and irradiation time of the plurality of estimation transmitting antennas the same as the irradiation order and irradiation time of the plurality of transmitting antennas.
9. 8. The estimation device according to 8, wherein the estimation image generation means generates the estimation image based on a part of the signals of the reflected wave received by the estimation electromagnetic wave transmitting / receiving means.
10. 9. The estimation device according to 9, wherein the estimation image generation means randomly selects a signal to be used for generating the estimation image from the signals of the reflected wave received by the estimation electromagnetic wave transmission / reception means.
11. The estimation image generation means includes the signal of the reflected wave based on the electromagnetic wave radiated from the estimation transmission antenna selected from the signals of the reflected wave received by the estimation electromagnetic wave transmitting / receiving means, and the signal of the reflected wave based on the electromagnetic wave radiated from the estimation transmitting antenna. 9. The estimation device according to 9, wherein a signal obtained by removing at least one of the reflected wave signals received by the selected estimation receiving antenna is selected as a signal to be used for generating the estimation image.
12. The estimation electromagnetic wave transmitting / receiving means irradiates the electromagnetic wave from the estimation transmitting antenna while changing the frequency.
The estimation image generation means selects a signal obtained by excluding a signal having a selected frequency from the reflected wave signals received by the estimation electromagnetic wave transmission / reception means as a signal to be used for generating the estimation image. 9 or 11 according to the estimation device.
13. The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The estimation electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of estimation transmitting antennas, receives reflected waves by the plurality of estimation receiving antennas, and receives reflected waves.
The number of the estimation transmitting antennas that irradiate the electromagnetic waves is smaller than the number of the transmitting antennas that irradiate the electromagnetic waves.
7. The estimation device according to 7, wherein the number of the estimation receiving antennas that receive the reflected waves is smaller than the number of the receiving antennas that receive the reflected waves.
14. The distance between the estimation transmitting antennas located at both ends in the left-right direction of the plurality of estimation transmitting antennas and the distance between the transmitting antennas located at both ends in the left-right direction among the plurality of transmitting antennas are set. 13. The estimation device according to 13.
15. The distance between the estimation receiving antennas located at both ends in the left-right direction of the plurality of estimation receiving antennas and the distance between the receiving antennas located at both ends in the left-right direction among the plurality of receiving antennas are set. 13. The estimation device according to 13 or 14, which is the same.
16. The computer
Irradiate electromagnetic waves from the transmitting antenna, receive reflected waves at the receiving antenna,
Based on the received signal of the reflected wave, a label determination image is generated.
A learning image is generated based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
The label is determined based on the label determination image, and the label is determined.
A processing method for generating teacher data in which the learning image and the label are associated with each other and storing the teacher data in the teacher data storage means.
17. The computer
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
The estimation model generated by the processing apparatus having the above is stored.
Electromagnetic waves are emitted from the estimation transmitting antenna, and the reflected wave is received by the estimation receiving antenna.
An estimation image is generated based on the received signal of the reflected wave, and the image is generated.
An estimation method for estimating an object included in the estimation image based on the estimation image and the estimation model.

Claims

Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
Processing equipment with.
The processing device according to claim 1, further comprising an estimation model generation means for generating an estimation model by machine learning based on the teacher data stored in the teacher data storage means.
The label determination image generation means
A means for outputting the label determination image and
A means for accepting the user input of the label of the output image for determining the label, and
The processing apparatus according to claim 1 or 2.
The process according to any one of claims 1 to 3, wherein the learning image generation means randomly selects a signal to be used for generating the learning image from the received signals of the reflected wave. Device.
The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The learning image generation means receives the signal of the reflected wave based on the electromagnetic wave emitted from the selected transmitting antenna from the received signals of the reflected wave, and the signal received by the selected receiving antenna. The processing apparatus according to any one of claims 1 to 3, wherein a signal excluding at least one of the reflected wave signals is selected as a signal to be used for generating the learning image.
The electromagnetic wave transmitting / receiving means irradiates the electromagnetic wave from the transmitting antenna while changing the frequency.
The learning image generating means selects a signal obtained by excluding a signal having a selected frequency from the received signals of the reflected wave as a signal to be used for generating the learning image, according to claims 1 to 3 and 5. The processing apparatus according to any one of 5.
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
An estimation model storage means for storing the estimation model generated by the processing apparatus having the
An estimation electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the estimation transmission antenna and receives reflected waves at the estimation reception antenna.
An estimation image generation means that generates an estimation image based on the received signal of the reflected wave, and
An estimation means for estimating an object included in the estimation image based on the estimation image and the estimation model, and
Estimator with.
The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The estimation electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of estimation transmitting antennas, receives reflected waves by the plurality of estimation receiving antennas, and receives reflected waves.
The electromagnetic wave transmitting / receiving means and the estimation electromagnetic wave transmitting / receiving means have the same configuration.
The estimation device according to claim 7, further comprising a parameter setting means for making the irradiation order and irradiation time of the plurality of estimation transmitting antennas the same as the irradiation order and irradiation time of the plurality of transmitting antennas.
The estimation device according to claim 8, wherein the estimation image generation means generates the estimation image based on a part of the signals of the reflected wave received by the estimation electromagnetic wave transmitting / receiving means.
The estimation according to claim 9, wherein the estimation image generation means randomly selects a signal to be used for generating the estimation image from the signals of the reflected wave received by the estimation electromagnetic wave transmission / reception means. Device.
The estimation image generation means includes a signal of the reflected wave based on the electromagnetic wave radiated from the estimation transmission antenna selected from the signals of the reflected wave received by the estimation electromagnetic wave transmitting / receiving means, and a signal of the reflected wave based on the electromagnetic wave radiated from the estimation transmitting antenna. The estimation device according to claim 9, wherein a signal excluding at least one of the reflected wave signals received by the selected estimation receiving antenna is selected as a signal to be used for generating the estimation image.
The estimation electromagnetic wave transmitting / receiving means irradiates the electromagnetic wave from the estimation transmitting antenna while changing the frequency.
The estimation image generation means selects a signal obtained by excluding a signal having a selected frequency from the reflected wave signals received by the estimation electromagnetic wave transmission / reception means as a signal to be used for generating the estimation image. The estimation device according to claim 9 or 11.
The electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of transmitting antennas, receives reflected waves at the plurality of receiving antennas, and receives the reflected waves.
The estimation electromagnetic wave transmitting / receiving means irradiates electromagnetic waves from the plurality of estimation transmitting antennas, receives reflected waves by the plurality of estimation receiving antennas, and receives reflected waves.
The number of the estimation transmitting antennas that irradiate the electromagnetic waves is smaller than the number of the transmitting antennas that irradiate the electromagnetic waves.
The estimation device according to claim 7, wherein the number of the estimation receiving antennas that receive the reflected waves is smaller than the number of the receiving antennas that receive the reflected waves.
The distance between the estimation transmitting antennas located at both ends in the left-right direction of the plurality of estimation transmitting antennas and the distance between the transmitting antennas located at both ends in the left-right direction among the plurality of transmitting antennas are set. The estimation device according to claim 13, which is the same.
The distance between the estimation receiving antennas located at both ends in the left-right direction of the plurality of estimation receiving antennas and the distance between the receiving antennas located at both ends in the left-right direction among the plurality of receiving antennas are set. The estimation device according to claim 13 or 14, which is the same.
The computer
Irradiate electromagnetic waves from the transmitting antenna, receive reflected waves at the receiving antenna,
Based on the received signal of the reflected wave, a label determination image is generated.
A learning image is generated based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
The label is determined based on the label determination image, and the label is determined.
A processing method for generating teacher data in which the learning image and the label are associated with each other and storing the teacher data in the teacher data storage means.
The computer
Electromagnetic wave transmission / reception means that irradiates electromagnetic waves from the transmitting antenna and receives reflected waves at the receiving antenna.
A label determination image generation means for generating a label determination image based on the received signal of the reflected wave, and a label determination image generation means.
A learning image generation means that generates a learning image based on a signal that is a part of the received signal of the reflected wave and is less than the signal used for generating the label determination image.
A label determining means for determining a label based on the label determining image, and
A teacher data generation means that generates teacher data in which the learning image and the label are associated with each other and stores the teacher data in the teacher data storage means.
An estimation model generation means that generates an estimation model by machine learning based on the teacher data stored by the teacher data storage means.
The estimation model generated by the processing apparatus having the above is stored.
Electromagnetic waves are emitted from the estimation transmitting antenna, and the reflected wave is received by the estimation receiving antenna.
An estimation image is generated based on the received signal of the reflected wave, and the image is generated.
An estimation method for estimating an object included in the estimation image based on the estimation image and the estimation model.